Discussion of Engle-Granger output as a whole and on variables

percent of the monthly movements in the Norwegian krone exchange rate in the period 2001-2015. In the first step of the two-step Engle-Granger procedure we found proof in favor of cointegration meaning that the long-run relation relationship can be interpreted.

We find all our explanatory variables to be significant in the long-term solution for the exchange rate. In the second step we try to determine the short-run dynamic in the model as well as how swiftly the krone exchange rate moves towards long-run equilibrium when it deviates from this. The coefficient on the error term informs us of the speed of

correction. This coefficient should be negative and have a value between -1 and 0, if not, a longterm equilibrium level will never be achieved. In our model the coefficient is -0.073 which means that 7.3 percent of a deviation of 1 percent from the long-term equilibrium adjusts back each month. The error term is significant at a 90% level and we claim the adjustment to be valid. The coefficients on our explanatory variables say something about the effect of a lagged variable change on the exchange rate. However none of the explanatory variables are significant within an acceptable confidence level and we cannot proceed to optimize our model by elimination of insignificant variables.

8.1.1 Oil price

In the Engle-Granger approach we find the oil price to be a significant variable at a 95%

confidence level in a long-run relation to the exchange rate. This is consistent with earlier research, both Bjørnstad & Jansen(2006) and Bernhardsen & Røysland(2000) found the oil price to play a role in the movements of the Norwegian krone. (o) is measured as the natural logarithm of oil price, thus we can interpret the result as elasticity. Stated in Table

52 11, a negative coefficient on (o) implies that a sustained increase in the oil price of 1 percent will lead to a real appreciation of the krone of 0.05 percent. Hence, having a negative relationship with the exchange rate as we pay fewer kroner per euro. Although there is uncertainty associated with the exact relationship between the oil price and NOK/euro exchange rate, the effect is consistent with what we would expect from theory.

We also find the coefficient to be negative when testing for the short-run relation, represented in Table 16. The coefficient on a lagged change in (o), (∆o_l1) states that an effect of a 1 percent change in (o) in the previous period provides a negative effect of 0.01 percent in a change in the exchange rate. In our estimation this is interpreted as an appreciation of the NOK. However we cannot determine (∆o_l1) to have an effect on the exchange rate in the short-run deviation from the long-run equilibrium level, as the p-value claims it to be insignificant.

8.1.2 Norwegian CPI

The Norwegian CPI (P) level can be used to measure the inflation by reflecting the Norwegian price levels. Theory predicts that if price levels increase it will cause currency depreciation. While research states that more commonly, inflation will have a significant negative effect than a significant positive effect on an exchange rate.

In the first step of the Engle-Granger approach we claim the long-run relation valid considering the p-value. Finding the variable (P) significant and to have a positive sign.

Our model states that a 1 percent permanent increase in (P) will cause a depreciation of the krone of 1.3 percent in the long run, represented in Table 11. The estimated

coefficient has the right sign in view of what we would expect from the relationship between inflation and exchange. The positive sign is also consistent with research by Bjørnstad & Jansen (2006) and Bernhardsen & Røysland(2000).

In the second step of the Engle-Granger approach we try to determine the short-term effect of a lagged change in (P), (∆P_l1), on the exchange rate if there is a deviation from

53 the long-term equilibrium level. Our model predicts that a 1 percent increase in (P) the previous month affects the exchange rate with a negative relation of 0.3 percent.

Considering our estimation this is understood as an appreciation of the NOK. The effect is represented in Table 16. We estimate the coefficient of (∆P_l1) to be insignificant and cannot find proof of the short-run effect. If (∆P_l1) was significant the adjustment of the Norwegian CPI to the equilibrium level exchange rate is a process that is slow and there is a great difference in the long and short-run coefficients.

8.1.3 Euro area harmonized CPI

The variable euro area harmonized CPI (P*) can be a measure of inflation in euro currency countries. In our model this is what we try to depict. From theory we believe that a decrease in the foreign price level will cause the exchange rate to decrease in value causing the NOK to appreciate against the euro in the long run.

From the first step in the Engle-Granger approach, as stated in Table 11, we can

determine the variable (P*) significant in a long-run relationship. This means that there is a causal relationship between (P*) and the exchange rate. Determining (P*) significant suggests that a sustained increase in (P*) of 1 percent will lead to a real appreciation of the Norwegian krone of 3.3 percent. This is consistent with findings in Bernhardsen &

Røisland(2000). We would expect an increase in the foreign index to lead the exchange rate to decrease in value causing the NOK to appreciate against the euro.

The variable appears in Table 16, in the output on the second step in the Engle-Granger approach, to have a negative coefficient. A negative coefficient implies an appreciation of the NOK. This is also harmonized with the beliefs originating from theory. The p-value states that the variable is not significant at a 95 % confidence level, leading us to declare it insignificant. No evidence of a 1 percent change in (P*) in the former period having a negative effect of 0.3 percent in the short run on the exchange rate is proved.

54 8.1.4 Interest rate differential (Nibor-Euribor spread)

The purpose of including the Nibor-Euribor spread (R-R*) was to enhance the interest rate differential effect on the exchange rate. Suggested from theory, if the 3-month money market rates differ, an increase in interest rates domestically, will lead to an appreciation of the Norwegian krone.

From the first step in the Engle-Granger approach we estimate the interest rate

differential to be significant in a long-run relationship. Bjørnstad & Jansen (2006) also find support of this. From the coefficient stated in Table 11, a 1 percent permanent increase in the (R-R*) will cause a real appreciation of the krone of 2.9 percent. This symbolizes that if there is a permanently higher increase in the domestic interest rates relative to the euro area the krone will increase its purchasing power compared to the euro. This consists with our beliefs from research by Kloster, Lokshall & Røisland (2003) about (R-R*) effect on the NOK/euro exchange rate.

We test the interest differentials short-run effect in the second step of the Engle-Granger approach. Here we find that a 1 percent increase in the lagged change of the interest rate differential (∆(R-R*)_l1) is shown to have a negative effect on the change in the

exchange rate with 0.02 percent. The negative coefficient entails and appreciation of the NOK. This is however not significant in the results from Table 16. We have no proof a short-run relation of the previous periods effect on the current exchange rate.

8.1.5 Norwegian balance of trade

Norway exports more than it imports, which means that there is a high demand for its goods and thus for its currency. When demand is high, prices rise, hence, the currency should appreciate. This is the effect that we wished to prove when including this variable.

In the first step of the Engle-Granger approach we find the explanatory variable balance of trade (x) to be significant in a long-term relation to the exchange rate. The variable has a positive coefficient, which implies that an increase in the balance of trade will affect the

55 Norwegian krone value negatively. This consists with findings in Lane & Milesi-Feretti (2002), they also find evidence of the balance of trade having a negative long-term relation to the exchange rate.(x) is measured as the natural logarithm of the Norwegian balance of trade, thus we can interpret the regression output on (7.1) as an elasticity.

Stated in Table 11, a 1 percent increase in the balance of trade with euro using countries will lead to a depreciation of the NOK of 0.02 percent. The positive coefficient is not what we would expect regarding theory. We do not find the effect of high demand of export increasing the currency demanded and thereby its value.

Referring to the second step in Engle-Granger we find the balance of trade coefficient in Table 16 to be negative and confirming our beliefs in reference to theory. The effect is opposite the results in the first step. A lagged first difference of (x), (∆x_l1) states that an effect of a 1 percent change in (x) in the previous period provides a negative effect of 0.05 percent change in the exchange rate. The negative coefficients suggests and appreciation of the NOK. The p-value is 0.35 and we cannot claim the effect valid.

Thereby we cannot prove a short-run relation if there is a deviation from the long-run equilibrium level on the exchange rate.

8.1.6 Unemployment rate

Our choice to include this variable in the model was to see if a productivity and stability measure would reflect in the Norwegian exchange rate. A rise in the unemployment rate is a sign of a weakening economy and can initiate cuts in interest rates. A cut in interest rates will increase demand and thus increase prices. Depreciation of the currency will normally be a consequence of low interest rates.

Testing for a long-run relationship in the first step of the Engle-Granger approach the variable unemployment rate appears to be significant on a 95% confidence level. A negative coefficient implies that a 1 percent increase in the unemployment rate will provide an appreciation of the krone with 2.2 percent in the long run as stated in Table 11. Our results from this model do not consist with our expectations of a depreciation of

56 the NOK. The negative coefficient implies Norwegians paying less per euro, and an appreciation of the NOK.

From the second step in the Engle-Granger approach, Table 16 cites that the coefficient has a positive sign and would confirm our belief about the variable in a short-run relation.

Although an insignificant variable, (∆U_l1) states that an effect of a 1 percent change in (U) in the previous period provides a positive effect of 0.06 percent on a change in the exchange rate. The positive coefficient would signify a depreciation of the NOK. We are also critical to our choice of variable for a productivity measure and believe that this variable does not estimate a good representation of the effects we wished to prove.